PERSPECTIVE

The Development of Computational Biologyin South Africa: Successes Achieved andLessons LearntNicola J. Mulder1*, Alan Christoffels2, Tulio de Oliveira3, Junaid Gamieldien2,Scott Hazelhurst4, Fourie Joubert5, Judit Kumuthini6, Ché S. Pillay7, Jacky L. Snoep8,Özlem Tastan Bishop9, Nicki Tiffin2

1 Computational Biology Division, Department of Integrative Biomedical Sciences, Institute of InfectiousDisease and Molecular Medicine, Faculty of Health Sciences, University of Cape Town, Cape Town, SouthAfrica, 2 South African National Bioinformatics Institute/Medical Research Council of South AfricaBioinformatics Unit, University of theWestern Cape, Bellville, South Africa, 3 Africa Centre for Health andPopulation Studies, School of Laboratory Medicine and Medical Sciences, University of KwaZulu-Natal,Durban, South Africa, 4 School of Electrical & Information Engineering, and Sydney Brenner Institute forMolecular Bioscience, University of theWitwatersrand, Johannesburg, South Africa, 5 Centre forBioinformatics and Computational Biology, University of Pretoria, Pretoria, South Africa, 6 Centre forProteomic and Genomic Research, Cape Town, South Africa, 7 School of Life Sciences, University ofKwaZulu-Natal, Pietermaritzburg, South Africa, 8 Department of Biochemistry, Stellenbosch University,Stellenbosch, South Africa, 9 Research Unit in Bioinformatics (RUBi), Department of Biochemistry andMicrobiology, Rhodes University, Grahamstown, South Africa

* nicola.mulder@uct.ac.za

AbstractBioinformatics is now a critical skill in many research and commercial environments as bio-

logical data are increasing in both size and complexity. South African researchers recog-

nized this need in the mid-1990s and responded by working with the government as well as

international bodies to develop initiatives to build bioinformatics capacity in the country. Sig-

nificant injections of support from these bodies provided a springboard for the establishment

of computational biology units at multiple universities throughout the country, which took on

teaching, basic research and support roles. Several challenges were encountered, for

example with unreliability of funding, lack of skills, and lack of infrastructure. However, the

bioinformatics community worked together to overcome these, and South Africa is now

arguably the leading country in bioinformatics on the African continent. Here we discuss

how the discipline developed in the country, highlighting the challenges, successes, and

lessons learnt.

IntroductionThe fields of bioinformatics and computational biology have grown in importance as drivers ofresearch in the life sciences as evidenced by the increasing number of journals and interna-tional conferences dedicated to these fields. The need for skills in these areas has grownaccordingly in South Africa, as researchers are more frequently awarded grants for projects

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 1 / 15

OPEN ACCESS

Citation: Mulder NJ, Christoffels A, de Oliveira T,Gamieldien J, Hazelhurst S, Joubert F, et al. (2016)The Development of Computational Biology in SouthAfrica: Successes Achieved and Lessons Learnt.PLoS Comput Biol 12(2): e1004395. doi:10.1371/journal.pcbi.1004395

Editor: Christos A. Ouzounis, Hellas, GREECE

Published: February 4, 2016

Copyright: © 2016 Mulder et al. This is an openaccess article distributed under the terms of theCreative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in anymedium, provided the original author and source arecredited.

Funding: Tulio de Oliveira’s research and databasesare supported by a Flagship grant from the MedicalResearch Council (MRC) of the Republic of SouthAfrica (MRC-RFA-UFSP-01-2013/UKZN HIVEPI).H3ABioNet is funded by the National Institutes ofHealth Common Fund under grant numberU41HG006941. The funders had no role in studydesign, data collection and analysis, decision topublish, or preparation of the manuscript.

Competing Interests: The authors have declaredthat no competing interests exist.

generating large-scale biological data. Local scientists are generating large and varied datasetsincluding next-generation sequencing (NGS; genomic, transcriptomic, and metagenomic),proteomic data, and other data, coupled with rich phenotypic datasets, especially large patientand surveillance cohorts. These research groups seldom have embedded data analysts, so theyturn to bioinformatics groups for support. Fortunately, there are well-established bioinformat-ics and computational biology groups at many of the universities in South Africa, reflecting thesignificant resources committed to bioinformatics development in the country since the 1990s.In this article, we outline the growth of bioinformatics in South Africa, highlighting the chal-lenges as well as the impact of government support on the development of the field. Where rel-evant, we also describe lessons learnt.

Origins: History and Development of Bioinformatics in South AfricaSouth Africa is a middle-income country (the World Bank’s 2013 estimate of GDP [GrossDomestic Product] per capita at purchasing power parity was US$12,530), with significantsocial inequality. While there are resources to develop new scientific fields, demands on gov-ernment to meet immediate social needs makes neither organic growth nor strategic invest-ment a given. Following the Second World War, state policy and government funding built astrong South African tradition of scientific research. Despite areas of great strength and inter-national leadership, however, there were many neglected areas, policy challenges, and a frag-mented science and innovation landscape across both the public and private sector [1–3]. Theending of sanctions and the accession of South Africa to the World Trade Organizationreduced protection and state subsidy to some areas of research that had previously been of stra-tegic importance (e.g., some areas of military research) [3].

Bioinformatics development post-1994 fell within a research and development (R&D) pol-icy adopted by the new democratic government to address the legacies of apartheid and eco-nomic needs. New science and research policies aimed to develop a coherent approach tobuilding research infrastructure and innovation. Blue-skies research was valued particularly inareas of competitive advantage such as palaeoanthropology, astronomy, and Antarctic studies.A key part of the new research policy including the national biotechnology strategy, however,explicitly aimed to jump-start the economy and address the serious economic challenges left bythe apartheid. An ongoing challenge is to address skewed demographics of researchers and thehigher education sector that arose from racial inequalities of the apartheid, in which the blackmajority was largely excluded from science [3]. This apartheid legacy is still apparent two gen-erations later in the failure of the school system to generate enough school completers for thescience-based careers that must be part of the growing economy. This is particularly true inareas like bioinformatics, where a strong mathematics training is required [3,4].

Early phasesThe development of bioinformatics in South Africa and on the African continent can be tracedback to 1996 whenWinston Hide founded the South African National Bioinformatics Institute(SANBI) on the University of the Western Cape (UWC) campus. South Africa’s NationalResearch Foundation (NRF) provided the initial funding to establish SANBI, and this wassupplemented with funding from what was then Glaxo Wellcome and the United StatesDepartment of Energy. The following two years saw investment from the South African tele-communications company, Telkom. During this time, the first four bioinformatics PhD stu-dents were enrolled at SANBI, and by 2000, South Africa’s vision for expanding bioinformaticswas realized with the establishment of the South African Medical Research Council Bioinfor-matics Capacity Development Unit at SANBI with committed funding for 2000–2015. In 2003,

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 2 / 15

a regional bioinformatics training center under the umbrella of the World Health Organiza-tion’s Tropical Disease Research (TDR) Program was established at SANBI. This program pro-vided a bioinformatics training mechanism for African researchers–a program thatsubsequently informed training within the International Glossina Genome Initiative [5].

Government support for biotechnology and bioinformatics was formalized in the 2001Green Paper on the Biotechnology Strategy [6], which highlighted the importance of humancapacity development and designated bioinformatics eligible for critical skills work permitapplications. Strategic and practical support continued, with ongoing updates of state policies(e.g., [7,8]). Several universities established nascent bioinformatics groups and lobbied collec-tively, resulting in an influential call to action published in the South African Medical Journal[9].

The National Bioinformatics NetworkThe first formal call for establishing a nationwide bioinformatics infrastructure came from theSouth African Medical Research Council in 2002, with funding from the national Departmentof Arts, Culture, Science, and Technology and later the Department of Science and Technology(DST). While the amount committed was modest, this catalyzed meetings of the bioinformaticsstakeholders in South Africa and subsequent formation of a group of scientists promoting theinterests of bioinformatics nationally. Shortly thereafter, the DST announced an initiative toestablish Biotechnology Regional Innovation Centers (BRICs), including a semiautonomousNational Bioinformatics Network (NBN). Planning for the NBN took place during 2002, andthe NBN was given a mandate to generate bioinformatics capability and infrastructure and actas a service provider, with specific emphasis on its role in supporting the BRICs. Calls for pro-posals to set up nodes for the network and for research projects were received in 2003, and aDirector was appointed, followed by administrative, technical and scientific staff. Initially, fivenodes were funded at Universities across South Africa, with two further nodes added at a laterstage. Node funding covered staff appointments, equipment purchases, general running costs,student bursaries, training courses, and research projects. This was followed by five furtherannual rounds of substantial funding. The NBN funding ended in 2008, and the NBN as anentity was closed in 2009. The NBN initiative vastly expanded bioinformatics capacity in SouthAfrica through the funding of staff positions and through the training of postgraduate students,as well as through extensively taught short courses for researchers. These activities all fell out-side of traditional university commitments. The investment in infrastructure provided a verysignificant boost to the scope of projects that could be executed and the level of bioinformaticsresearch ongoing.

In retrospect, some of the lessons learned from the NBN initiative may be useful for others.These include:

• Splitting of funds into noncompetitive funding for training and initial infrastructure andcompetitive funding for research-fostered cooperation between bioinformatics groups, whileencouraging healthy competition for funding of research projects.

• The use of an independent, international Scientific Advisory Board (SAB) to mediate keyfunding decisions ensured quality and nonpolitical funding decisions, particularly at theearly phases when the South African bioinformatics community was too small to supportindependent review at a national level.

• Although the state identified bioinformatics as key to supporting health priorities within thebiotech industry, the need to build general infrastructure and research capacity was also rec-ognized, and the NBN was given wide autonomy for the type of research funded.

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 3 / 15

• A very complex governance structure included the NBN Trust Committee, the SAB, and anexecutive, and this led to a breakdown in accountability when problems did occur.

• Whilst the dedicated funding was very beneficial, the process was poorly integrated into thenational funding system and researchers had to source funding from different entities tocover different aspects of a single project. For example, funding for related laboratory workhad to be sourced under different initiatives.

Following closure of the NBN initiative, the DST formed a steering committee for Bioinfor-matics in 2009, which facilitated bioinformatics research and fellowships disseminated throughthe NRF, commencing in 2010 and repeated every two years. Funding for a national Bioinfor-matics Service Platform (BSP) was granted in 2013, and the appointment of a coordinator iscurrently underway (discussed further below). Further support for bioinformatics research hasbeen provided through the DST/NRF South African Research Chairs Initiative (SARChI), andthree chairs are currently supported in the broad bioinformatics area (Bioinformatics and Pub-lic Health Genomics, Bioinformatics of African Populations, Mechanistic Modelling of Healthand Epidemiology). Fig 1 provides a summary timeline of the major events impacting thedevelopment of bioinformatics in South Africa.

International influences on the development of bioinformatics in SouthAfricaSeveral international initiatives influenced the development of bioinformatics in South Africa.In the early 1990s, the European Molecular Biology Network (http://www.embnet.org/) estab-lished the first bioinformatics network in Africa, with nodes similar to EMBNet Nodes inEurope. The first African Node was set up at SANBI, followed by many others (http://www.embnet.org/about/nodes). The World Health Organization’s TDR also provided funding tosupport establishment of an African Regional Training Centre for Bioinformatics and AppliedGenomics in South Africa. As part of this activity, many training workshops were presentedacross Africa. Another approach to set up a large network within Africa, with leadership inSouth Africa, was led by the Centre National de la Recherche Scientifique (CNRS) in collabora-tion with NEPAD. This network funded the start of the ASBCB (African Society for Bioinfor-matics and Computational Biology) Bioinformatics conferences (ASBCB later joined forceswith the International Society for Bioinformatics and Computational Biology [ISCB] to runthese conferences) and workshops and mapped the bioinformatics structure in Africa (http://www2.lirmm.fr/france_afrique/20060708_NEPAD.pdf). These international activitieshelped to establish the initial bioinformatics network structure in South Africa, which in turnattracted national government recognition and commitment to advancing bioinformatics inthe country.

More recently, international initiatives are having a profound influence on South Africanbioinformatics. The NIH-funded H3ABioNet Pan African bioinformatics network (www.h3abionet.org) is part of the Human Heredity and Health in Africa (H3Africa: www.h3africa.org, discussed later) initiative [10] and is coordinated from the University of Cape Town(UCT) with six additional nodes in South Africa. This project is building bioinformatics capac-ity on the continent and, as a result, is boosting bioinformatics activities in South Africa. Theaward of funding for H3ABioNet in South Africa prompted the DST to commit funds for bio-informatics support personnel at the Centre for High Performance Computing (CHPC) inCape Town and to continue funding national bioinformatics courses for postgraduates.

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 4 / 15

Infrastructure Development for BioinformaticsOriginally, costs for bandwidth in South Africa were prohibitive and bandwidth was low, withSANBI having a relatively good connection initially at a dedicated 1 Mbps in 2003. The lack ofbandwidth posed huge challenges to the development of bioinformatics, and especially toaccessing large international databases. The situation slowly improved in a disjointed fashionin different institutions, but progressed substantially only from 2006, with the establishment bythe DST of the South African National Research Network (SANReN) and TENET (The Ter-tiary Education and Research Network of South Africa; an Internet Service Provider run by theuniversities). SANReN built a national fiber backbone with speeds up to 10 Gbps. TENEToperated the national network and procured international bandwidth [11]. Currently, TENEThas a 10 Gbps link to Amsterdam on the SEACOM (submarine cable operator) cable and a 10Gbps link on the WACS (West Africa Cable System) cable to London, and manages transit tointernational networks including Géant (http://www.tenet.ac.za). This has led to substantialimprovements in bandwidth availability for reduced costs. Although network bandwidthremains limited compared with many other countries and the high latency caused by physicaldistance imposes challenges, bandwidth has now reached the point where it is feasible to trans-fer terabyte-size datasets internationally.

Fig 1. The policy, infrastructure, training, and funding landscape of computational biology in South Africa (2000–present).Government policy oncomputational biology was driven largely by the DST (above timeline arrow), while bioinformatics training courses have been a constant feature of thislandscape since 2003. There was a major period of infrastructure investment from 2002–2007, from the NBN, but national funding has now become availableonly every two years. National bioinformatics conferences also occur in two-year intervals (details in text).

doi:10.1371/journal.pcbi.1004395.g001

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 5 / 15

The first major infrastructure for bioinformatics computing was Silicon Graphics SMP sys-tems installed at SANBI in 1997, followed by the installation of Cray SV1 in 2002. Subse-quently, the NBN funded the installation of some high-end SMP systems and clusters in theperiod from 2003–2009. Many of the active bioinformatics groups run their own clusters, andcarry the bulk of the workload of bioinformatics in South Africa. However, national facilitieshave been a very valuable resource for the community. An Intel-funded cluster for Life Scienceswas set up at the Council for Scientific and Industrial Research in 2005 for use by the greaterbioinformatics community. The CHPC was established in 2007 with the installation of an IBMcluster with 160 cores and two IBM p690 systems. Later acquisitions include IBM BlueGene Pin 2008 and a Sun M9000 SMP system with 2 TB of RAM in 2009. The current cluster is a het-erogeneous cluster of Oracle and Dell machines with a peak performance of 61 teraflops andover 7,000 cores and 480 TB of external storage [11]. A significant upgrade is currently plannedwith some dedicated resources for bioinformatics. A new DST-funded Data Intensive ResearchInitiative of South Africa is being established to support large-scale data storage. There havealso been a number of examples of the use of public cloud facilities.

In examining the successes and shortcomings to date, there are two key lessons: (1) lack ofspecialized skills (bioinformaticists, bioinformatics data analysts, systems administrators, andspecialist programmers) has been more of a stumbling block than access to hardware resources;and (2) the most expensive equipment does not necessarily make the biggest contribution,even when the upfront capital costs are very low (e.g., through a donation). Thus, initiativessuch as the CHPC appointing two dedicated specialized bioinformatics staff members to sup-port the bioinformatics community are particularly positive.

Bioinformatics Education and TrainingIn South Africa, bioinformatics teaching is mainly offered at the postgraduate level, thoughsome universities offer short undergraduate modules. Students accepted to postgraduate bioin-formatics degree programs have a variety of backgrounds, including biological sciences, com-puter science, mathematics, physics, and statistics. Although there are only a handful ofuniversities giving formal qualifications in bioinformatics in the country, a number of otheruniversities have projects in which bioinformatics is either the full focus or a component of theresearch (See Table 1 for programs available and students trained). Students from these univer-sities are supported by attending national bioinformatics courses or training activities, byattending bioinformatics modules at their own university, or through collaboration with bioin-formatics groups.

National bioinformatics coursesBioinformatics was not traditionally part of undergraduate curricula in South Africa, so stu-dents registering for postgraduate degrees in bioinformatics tend to start with little formaltraining in the field. In response, the NBN developed joint courses compulsory for NBN-funded students that introduced them to a range of bioinformatics topics and developed theirprogramming and other technical skills. The first courses were run over two semesters of eightweeks each, but the length and curriculum has been refined over time to account for changingneeds, costs, and realistic length of time students can spend away from their research projects.Today, these courses run for seven weeks at the start of each academic year and are aimed atstudents registered for dissertation-only postgraduate bioinformatics degrees at South Africanuniversities. At the start, we relied heavily on international trainers, but now all trainers arebased within the country, a testament to the building of local training and skills capacity. Inaddition to exposure to a range of topics, computer science and mathematics students learn

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 6 / 15

basic molecular biology, while biology-trained students learn programming. All the studentsbenefit from networking opportunities and sharing skills and experience with their peers.These courses, which are supported financially by the DST, are extremely popular and oversub-scribed, so we are considering running two such courses in different geographical locationswhere training is needed, and in the future. The substantial number of applications from benchscientists to attend these courses highlights the current gap in the training of bioinformaticsusers.

Other bioinformatics training initiativesIn addition to these national courses for training bioinformaticists, different universities andthe NBN have run short specialized training courses funded by their own institutions, foreignfunding agencies, the NBN or directly by the DST. Some have hosted short courses organizedthrough collaborative grants that individual researchers have with other countries or throughfunds raised from initiatives such as the NRF Knowledge, Interchange, and Collaboration pro-gram. Many of these have focused on training biological scientists in the use of bioinformaticstools, but some have also been tailored towards extending the skills of bioinformaticists. Someexamples are provided below but are by no means comprehensive.

Several researchers at South African universities have hosted Ensembl and other roadshowson specific databases or tools, including Galaxy (http://galaxyproject.org/). They have also

Table 1. Bioinformatics degree programs offered at some of the South African universities. The numbers shown are for registrations or graduationswhere the content is bioinformatics. The key research areas of each institution are included.

University Start ofgroup

Area of research Graduates todate

Nelson Mandela Metropolitan University (NMMU) 2014 Structural bioinformatics of immune receptors

Rhodes University (RU), Research Unit in Bioinformatics(RUBi) http://rubi.ru.ac.za/

2003* Structural Bioinformatics, Comparative Genomics, ToolDevelopment, Database Development

2 Honors, 25 MSc(1 yr)**, 3 MSc,2 PhD

Stellenbosch University (US) Molecular Systems Biologygroup

2000 Computational Systems Biology; mechanistic modelling ofhealth and epidemiology; yeast, bacterial and plant systemsbiology; model database; model simulation software.

17 MSc, 9 PhD

UCT, Computational Biology Group http://www.cbio.uct.ac.za/

2003 Bioinformatics & Systems Biology of infectious diseases,population genetics and disease, development of newalgorithms and visualization for high-throughput biology

17 Honors,16 MSc, 9 PhD

University of Free State http://cbio.ufs.ac.za/ 2005 Epigenomics 4 MSc

University of KwaZulu Natal (UKZN) http://lifesciences.ukzn.ac.za/Staff/Biotechnology/biotech_staff/Biotech_pmb/pillayc.aspx; http://www.bioafrica.net

2009 Redox Systems Biology, HIV and tuberculosis (TB) drugresistance, evolutionary biology, development of onlinedatabases and bioinformatics tools

8 MSc, 5 PhD

University of Pretoria (UP), Bioinformatics andComputational Biology Unit http://www.up.ac.za/centre-for-bioinformatics-and-computational-biology

2002 Cancer Genomics, Bacterial Genomics 32 Honors,29 MSc, 10 PhD

UWC, SANBI http://www.sanbi.ac.za 1996 Genome assembly, identification of human disease genesusing NGS, transcriptomics, network biology, viralgenomics and phylogenetics, pathogen genomics and drugresistance, evolutionary biology, in-silico drug leaddiscovery, semantic database development.

26 MSc, 29 PhD

University of Witwatersrand (Wits) http://www.bioinf.wits.ac.za

2003 Novel algorithm development, genome-wide associationstudy (GWAS) and population structure, NGS analysis forstudying gene/disease interaction

2 Honors, 6 MSc,2 PhD

*Activities suspended 2005–2009

**Machanick and Tastan Bishop 2015 [12]

doi:10.1371/journal.pcbi.1004395.t001

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 7 / 15

organized training around specialized topics or techniques, such as microarray, proteomics, orNGS data analysis. These are initiated through personal contacts, and funding sources vary.While such short-course training is not coordinated in South Africa, in many cases fundinghas been shared, particularly where the course runs in multiple locations in the country. Thisprovides a cost-effective means of training many people in each region. These courses havefocused mainly on training researchers in South Africa, but several courses have been hosted inthe country and targeted to those both internal and external to the country. For example,SANBI hosted annual WHO TDR workshops for African scientists, and the results of one ofthese workshops are included in a recent Science publication [13]. UKZN and MedicalResearch Council in Durban have also served as a hub of education and training on bioinfor-matics, training people both locally and abroad. International organizations that have had atraining presence in the country are the ICGEB (International Centre for Genetic Engineeringand Biotechnology), Wellcome Trust, EMBO (European Molecular Biology Organization),EMBNET, and the NIH, through H3ABioNet. Training workshops were also held in 2011 inCape Town as part of the ISCB Africa ASBCB conference on bioinformatics [14]. In additionto providing training courses, several South African academics are actively involved in GOB-LET (Global Organization for Bioinformatics Learning, Education and Training) [15].

The DST, in addition to supporting national bioinformatics courses, has also made provi-sion for financial support of necessary skills development and training needs, in the form ofworkshops and conferences as required by the bioinformatics community. It extends its train-ing initiative through international cooperation partnership agreements, e.g., EMBO andCooperation in Science and Technology (COST) training, as well as a number of other DST-driven platforms and Centers of Competence to enhance the bioeconomy of South Africa [8].Also, as part of the Bioinformatics and Functional Genomics (BFG) program, which is admin-istered through the NRF to fund bioinformatics research, the DST supports the developmentof human capacity and the creation of BFG skills for bio-based innovation and technology ini-tiatives. Through this, South Africa is creating a growing pool of postgraduate studentsequipped to support bioeconomy developments, solutions, and knowledge relevant to theSouth African biotechnology industry, and national initiatives such as the Technology Innova-tion Agency (TIA), Centers of Excellence, and Research Chairs. It also aims to promote collab-oration between academic institutions, science councils, and industry in South Africa, thuslinking bioinformatics education to other national initiatives. There are various other trainingmechanisms to complement as well as to strengthen the training needed to bridge the gapbetween the academia and industry. Examples include internships offered by the TIA, NRF,and H3ABioNet, train-the-trainer programs, and a Knowledge Transfer Program (KTP) by theCentre for Proteomic and Genomic Research (CPGR).

Development of a Bioinformatics Support Structure for Researchand IndustryBioinformatics capacity is essential in order to harness genomics research capacity, due to thelarge datasets generated and the complex analyses they require [16]. In 2008, the South AfricanDST launched a Ten-Year Innovation Plan, with a “Farmer to Pharma” component that aimedto strengthen the bioeconomy and grow biotechnology in the country (http://www.dst.gov.za/images/pdfs/The%20Ten-Year%20Plan%20for%20Science%20and%20Technology.pdf. Animportant goal of the DST in funding bioinformatics initiatives was to provide bioinformaticssupport to researchers in academia and industry, and many of the funded activities activelycontributed to this mission. Support included providing advice on and help with bioinformat-ics analyses, as well as developing new tools and resources and skills required locally. One of

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 8 / 15

the major focus areas in the bioeconomy strategy is health. Although there are many examplesof the application of bioinformatics to agriculture and other focus areas in SA, here we only dis-cuss how some of the specific bioinformatics needs in health-related fields are being addressed.

Bioinformatics in healthGenomics, biotechnology, and specifically bioinformatics were areas that were identified asrequisite for addressing the heavy burden of disease in South Africa, particularly diseases ofpoverty [17]. To date, medical genomics research in South Africa has focused almost exclu-sively on infectious diseases that include HIV, malaria and tuberculosis. Bioinformaticsapproaches are used to identify host, vector, and pathogen variation, and to identify variantsand biomarkers of pathogenicity and multidrug resistance. Genomic approaches have alsobeen applied to infectious disease control [18]. South Africa, however, is in the process of epi-demiological transition from a major disease burden of infectious to noncommunicable dis-eases [19,20]. Furthermore, the advent of NGS technologies, their increasing accessibility, anddecreasing costs have opened new avenues to understanding genotype–phenotype relation-ships in noncommunicable human disease [20], and the high genetic diversity of sub-SaharanAfrican populations positions South Africa to conduct groundbreaking genomics researchtowards understanding genetic factors underlying human disease [21].

H3Africa initiative. In 2012, human health genomics research in Africa received a signifi-cant boost through the H3Africa program [10] jointly funded by the National Institutes ofHealth (US) and the Wellcome Trust (UK). Out of over 20 funded projects in this program, tenare led from South Africa. Bioinformatics capacity for health genomics research has specificallybeen prioritized within H3Africa through the H3Africa Bioinformatics Network (H3ABioNet)mentioned previously. H3ABioNet [22] is building capacity by developing computing infra-structure, bioinformatics graduates, and training programs for genomics researchers. It isaddressing all computational aspects of genomics projects from providing support for patientdatabases, through full data analysis, to enabling submission to public repositories. This isbeing achieved through training and pooling of resources across the network partners todevelop a solid support structure.

HIV drug resistance databases. A good translational example of the usage of bioinformat-ics applications for health in South Africa was the setup of HIV drug resistance databases. Thescaling up of access to antiretroviral therapy (ART) in southern Africa had an extraordinaryeffect on extension of survival among those infected [23]. However, a major challenge is thedevelopment of drug resistance. Bioinformatics methods are widely used in developed coun-tries to identify drug resistance mutations in HIV patients on ART [24]. As part of a collabora-tive project between the Wellcome Trust Africa Centre at UKZN and the MRC, the twoleading drug resistance databases in the world, housed at Stanford University (Stanford, US)and the Rega Institute (Leuven, Belgium), were transferred to bioinformatics servers in SouthAfrica and made publicly available [25,26]. These databases were published in HIV Drug Resis-tance regional guidelines [27] and are implemented in the national treatment programs ofBotswana and South Africa [28].

South African Human Genome Program. The South African Human Genome Program(SAHGP) was launched in January 2011, with support from the DST. Its goals are to: (1)develop capacity for genomic research in southern Africa, (2) establish a sustainable resourcefor genomic research, and (3) translate the information and knowledge into improvements inhuman health [29]. Significant work has gone into the planning of the SAHGP, especially withregard to ethical, legal, and social issues, which are particularly important in our environment[30]. A pilot phase of the SAHGP is currently under way; 24 complete genomes have been

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 9 / 15

sequenced and are currently under analysis. This provides an important training opportunityfor enabling South African scientists to build capacity for large-scale sequence data analysis.

Computational systems biology for health. From its initiation, the NBN held a broadcomputational biology view of bioinformatics, and it included Computational Systems Biologyas one of its course topics. Systems biology formed part of research groups in several of the net-work nodes. In addition to training students in the field, systems biology software tools e.g., asimulation engine, PySCeS [31], and a curated model database JWS Online [32] were devel-oped and are publically available. The model database is linked to several international scien-tific journals and in large European systems biology projects as part of a data and modelmanagement structure. Many systems biology projects are applied in biotechnology or health-related projects; the recently started SARChI chair in “Mechanistic modelling of health andepidemiology” with a strong link to the South African Centre for Epidemiology and Mathemat-ical Analysis in Stellenbosch is an excellent example. The JWS Online database is funded vialarge international Systems Biology grants for which it is used as part of the data and modelmanagement team (e.g., SysMO I and II, and currently ERASysApp). The database is main-tained at Stellenbosch University (SUN) with mirror sites at the University of Manchester andthe Vrije Universiteit in Amsterdam. Funding to maintain the databases has been guaranteedfor at least ten years after the current funding round ends. The South African part of the initia-tive is funded via the long term SARCHI initiative.

Establishment of a National Bioinformatics Support PlatformImproved affordability and availability of high-throughput technologies to the South Africanscientific community has generated increasingly complex genomic datasets, with a concomi-tant increase in the need for bioinformatics support. Since the dissolution of the NBN, theexisting bioinformatics groups have generally been able to secure funding for research, but notfor service provision. Despite this, there has been an expectation from peers at their universitiesthat they should play a role in providing support to researchers. This puts them in a difficultsituation as salaries for support staff need to be covered and they should derive some benefitfrom their involvement. The analysis, however, doesn’t always lead to publications. In addition,this may address some of the needs of researchers located near competent bioinformaticsgroups, but not those of other researchers or the biotechnology community. In response, a pro-posal for a National BSP was developed by a Steering Committee, and subsequently funded bythe DST. The platform is currently being established and staffed, and will provide a bioinfor-matics support structure for researchers in the academic and commercial sectors throughoutthe country. The BSP will respond to service requests nationally based on expertise and man-aged by a central national helpdesk. Although centrally coordinated, BSP staff will be secondedto institutions who demonstrate significant and ongoing need for support. The BSP will alsotake over custodianship of national bioinformatics training and education that lies outside ofuniversity responsibilities. The hope is that the BSP will fill the current gap in bioinformaticsservice provision that has come to be expected by scientists.

Local and International LinksThe computational biology community in South Africa is continuing to promote local andinternational links through crossinstitutional grants from individuals, large networks, andmore formal organizations or societies. Increasingly, grants are awarded to multicenter net-works, for example, a BFG grant from the National Research Foundation of South Africa hasbeen awarded to partners at UWC, UCT, UP, SUN, University of the Witwatersrand, Univer-sity of Limpopo, and the Agricultural Research Council for the development of computational

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 10 / 15

resources to support analysis of Southern African human genomes. Separately, SUN and UWCare funded under this program to investigate genomic and proteomic determinants ofMyco-bacterium tuberculosis phenotypic characteristics. Collaboration across Africa is also increas-ing, for example the Medical Research Council of South Africa funds a Flagship project todevelop computational tools for tuberculosis research, with partners at UWC, UKZN, SUN,UCT, and the University of Benin. The South Africa–Kenya grant funds a program to charac-terize miRNA in disease vectors, with partners UWC and ICIPE in Kenya.

Outside Africa, there are many active collaborations, for example the Africa Centre forHealth and Population Studies at UKZN collaborates with the UCT Computational Biologygroup and the Swiss Institute of Bioinformatics (SIB) on the Swiss–South Africa grants andwith CNRS, Leuven Katholeic University, University of Manchester, and Erasmus MC H2020EC project (VIROGENESIS: developing of next generation tools for virus NGS data). Inanother example, the Computational Biology Group at UCT partnered with the European Bio-informatics Institute (EBI) and eight other institutions in Europe and the US on a EuropeanUnion (EU) FP7 project (contributing to the InterPro database), and has ties with several Euro-pean bioinformatics groups, including SIB, facilitated by Swiss–South Africa and Spain–SouthAfrica grants. Research from the group feeds directly into the Swiss-based TubercuList, one ofthe key reference databases forMycobacterium tuberculosis.

South African Society for Bioinformatics (SASBi)Local bioinformatics groups were adversely affected by the closing of the NBN, but the impactwent beyond the financial, since the NBN also provided a voice and a discussion platform forthe bioinformatics community in the country. In order to bring the bioinformatics communitytogether, the community recently established SASBi (http://sasbi.weebly.com/). The aim was tosupport newly established bioinformatics research groups, to enhance interactions among peo-ple related to bioinformatics, and to form a representative body to interact with government.An interim steering committee was formed to establish the national society, and this committeeproduced a draft constitution. The society was inaugurated on 11 September 2012 at a meetingduring the first joint conference of the South African Genetic Society (SAGS) and the bioinfor-matics community. The second joint SAGS/SASBi conference took place in September 2014.Both DST and BSP have provided financial support which made it possible to support 71 stu-dents to attend the conference. SASBi has interacted with the DST on a number of occasions,e.g. it conducted a national survey for bioinformatics needs, and discussed the BSP proposal.During the second joint conference, a SASBi Student Society was formed, and has already beenaffiliated with the international society (ISCB).

International societies and networksMany members of SASBi are also members of international societies such as the ASBCB andthe ISCB. At least four South African bioinformaticians have previously been or are currentlymembers of the ISCB Board of Directors, and one has served two terms as President of ASBCB.South Africa hosted one of the ISCB Africa ASBCB Bioinformatics conferences in Cape Townin 2011. The SASBi student society was affiliated with ISCB in March 2015, and the affiliationof SASBi with ISCB will be initiated this year now that the society has been in existence for aminimum of two years. Individual bioinformatics researchers also have numerous interna-tional collaborators both on the continent and abroad, some of which are described above. Anotable example of a large collaborative network is the previously mentioned H3ABioNet net-work, which links seven local bioinformatics groups with over thirty groups continent-wideand two in the US, and is led from South Africa. The project also links the bioinformatics

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 11 / 15

community with genomics researchers and clinicians in Africa through the wider H3Africaconsortium, which consists of over 350 members in 18 African countries.

The EMBNet has been a direct and indirect supporter of bioinformatics initiatives in SouthAfrica. They have, for example, provided numerous reviewers for national grant applications,identified partners for various successful and unsuccessful grant applications, sponsorednational conferences, and provided technical support for bioinformatics queries through theirmailing lists. They have run training courses in the country and recently facilitated the linkbetween South Africa and the EU COST action, SeqAhead (http://www.seqahead.eu/). SouthAfrica was the first African country to join the Action BM1006 called the NGS network. SeqA-head offered various workshops on NGS data and aimed to galvanize the implementation ofefficient workflows for NGS data storage, retrieval, sharing, and analysis.

ConclusionsBioinformatics in South Africa started with a single group at one South African university, butdeveloped rapidly through the injection of funds from the DST via the NBN initiative. TheNBN provided a springboard for the development of bioinformatics skills in postgraduate stu-dents and bench scientists, and these new initiatives are ensuring a continuation of thismomentum. Since then, the Bio-Innovation Unit of the DST has invested in several key bioin-formatics-based initiatives, including, the BFG program; the BSP; and the SAHGP. They havecontributed to a significant increase in bioinformatics outputs including trained postgraduatestudents, research publications (see Fig 2), patents, and bioinformatics tools and resources.There have also been downstream impacts, such as creation of new bioinformatics jobs, smallbusiness startups, and enhanced bioinformatics service delivery. Examples of these includeHealthQ Technologies (http://www.healthq.co), a start-up company originating at Stellenboschwith founding members from the computational systems biology group; and Electric Genetics,which developed StackPACK from EST (Expressed Sequence Tag) clustering and transcriptreconstruction software originating from SANBI, which was subsequently used by Affymetrixas part of its expression array design process. Electric Genetics also cosponsored and hosted

Fig 2. Relative increase in bioinformatics and computational biology publications from South Africacompared to publications in other subject areas. Fold change indicates relative increase in publicationnumbers compared to number of publications in 2001.

doi:10.1371/journal.pcbi.1004395.g002

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 12 / 15

the first ever two-part Open Bio Hackathon with O’Reilly, which was a major driver towardsthe first BioPerl publication. To demonstrate the impact on publications, we conducted asearch of PubMed (on 22 January 2015) using the following search parameters: [“subject area"]and "south africa"–"working in partnership with the European Bioinformatics Institute" and “.ac.za”, where “subject area” was [“bioinformatics” or “computational biology”], “chemistry,”“cardiology,” “geology,” “microbiology”. Patents and citations were excluded. We chose to sim-plify the search and use the broad terms bioinformatics and computational biology, as it ischallenging to identify all bioinformatics publications by journal or topic alone. We thereforeincluded the other subjects to demonstrate the relative differences in impact using broad sub-ject terms. It is evident that the number of bioinformatics-related publications in South Africahas increased substantially since 2001, with a sharper rise than other subjects over the sametime period.

Some of the key achievements of the South African bioinformatics community lie in train-ing, research, and infrastructural capacity development. The universities have collectivelytrained 130 Masters and 60 PhD students in bioinformatics, many of which have gone on toacademic positions or placements in prestigious institutes abroad, including Harvard and theEBI. Bioinformatics courses in South Africa are increasingly being taught by exclusively localtrainers, many of whom are frequently invited to give training courses in Africa and abroad.Several groups have contributed to or developed resources used by the international scientificcommunity, one example being the HIV drug resistance database. The strength and interna-tional recognition of the community is reflected in the participation of South African academ-ics in key positions in international societies, such as ISCB and ASBCB, partnerships withinternational research groups in successful funding proposals, and the increase in publicationsin international bioinformatics journals. One of the highlights, though, is the NIH award forH3ABioNet, which is one of the largest single awards to an African institution.

Throughout the history of the development of bioinformatics in SA, many lessons havebeen learnt, some of which are described above. The take-home message is that commitmentto bioinformatics capacity development by the government has been key to the success ofmany of the bioinformatics units in the country. The step-up provided by the NBN enabledgroups in multiple universities to hire dedicated bioinformatics staff members in an environ-ment where universities were not forthcoming in creating new positions. In some cases, thesuccess of the groups convinced the institution of the importance of bioinformatics and theneed to support it. The establishment of these bioinformatics units also facilitated the procure-ment of further funding from the NRF as well as international funding agencies, thus increas-ing outputs and enabling these units to become internationally competitive in the field.

AcknowledgmentsThe authors acknowledge the support of the South African Department of Science and Tech-nology and the National Research Foundation.

References1. Kahn M. The South African national system of innovation: From constructed crisis to constructed

advantage? Sci Public Policy. 2006; 33: 125–136. doi: 10.3152/147154306781779109

2. Kaplan D. Science and Technology Policy in South Africa Past Performance and Proposals for theFuture. Sci Technol Soc. 2008; 13: 95–122. doi: 10.1177/097172180701300104

3. Kahn MJ. Rhetoric and Change in Innovation Policy: The Case of South Africa. Sci Technol Soc. 2013;18: 189–211. doi: 10.1177/0971721813489447

4. Spaull N, Kotze J. Starting behind and staying behind in South Africa: The case of insurmountablelearning deficits in mathematics. Int J Educ Dev. 2015; 41: 13–24. doi: 10.1016/j.ijedudev.2015.01.002

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 13 / 15

5. Christoffels A, Masiga D, Berriman M, Lehane M, Touré Y, Aksoy S. International glossina genome ini-tiative 2004–2014: a driver for post-genomic era research on the African continent. PLOS Negl TropDis. 2014; 8: e3024. doi: 10.1371/journal.pntd.0003024 PMID: 25144472

6. DACST. A National Bioinformatics Strategy for South Africa, Department of Arts, Culture, Science andTechnology of the Republic of South Africa. 2001.

7. DST. Ten Year Strategy for Science and Technology. Department of Science and Technology of theRepublic of South Africa. 2008.

8. DST. The Bio-Economy Strategy. Department of Science and Technology of the Republic of SouthAfrica. 2014.

9. HideW, Mizrahi V, Venkatesh B, Brenner S, Simpson A, Blatch G, et al. A platform for genomics inSouth Africa. South Afr Med J Suid-Afr Tydskr Vir Geneeskd. 2001; 91: 1006–1007.

10. H3Africa Consortium. Research capacity. Enabling the genomic revolution in Africa. Science. 2014;344: 1346–1348. doi: 10.1126/science.1251546 PMID: 24948725

11. Petruccione F, Hazelhurst S, Host G, Lourens A, McIntyre C, Moore R. National Integrated Cyberinfras-tructure System. A framework of the establishment and Maintenance of a sustainable NICIS. Depart-ment of Science and Technology; 2013.

12. Machanick P, Tastan Bishop Ö. How to establish a bioinformatics postgraduate degree programme—acase study from South Africa. Brief Bioinform. 2015; 16: 346–354. doi: 10.1093/bib/bbu014 PMID:24794523

13. International Glossina Genome Initiative. Genome sequence of the tsetse fly (Glossina morsitans): vec-tor of African trypanosomiasis. Science. 2014; 344: 380–386. doi: 10.1126/science.1249656 PMID:24763584

14. De Villiers E, Kumuthini J, Bongcam-Rudloff E. ISCB Africa ASBCB Conference on Bioinformatics andeBioKit Workshop. EMBnet J. 2011; 17: 7–9.

15. Atwood TK, Bongcam-Rudloff E, Brazas ME, Corpas M, Gaudet P, Lewitter F, et al. GOBLET: TheGlobal Organisation for Bioinformatics Learning, Education and Training. PLoS Comput Biol. 2015; 11:e1004143. doi: 10.1371/journal.pcbi.1004143 PMID: 25856076

16. Daar AS, Thorsteinsdóttir H, Martin DK, Smith AC, Nast S, Singer PA. Top ten biotechnologies forimproving health in developing countries. Nat Genet. 2002; 32: 229–232. doi: 10.1038/ng1002-229PMID: 12355081

17. Al-Bader S, Frew SE, Essajee I, Liu VY, Daar AS, Singer PA. Small but tenacious: South Africa’s healthbiotech sector. 2009; 427–445.

18. OECD. Public Health in an Age of Genomics [Internet]. Paris: Organisation for Economic Co-operationand Development; 2013 Aug. http://www.oecd-ilibrary.org/content/workingpaper/5k424rdzj3bx-en

19. Mensah GA, Mayosi BM. The 2011 United Nations high-level meeting on non-communicable diseases:the Africa agenda calls for a 5-by-5 approach. 2013; 77–79.

20. Peprah E, WonkamA. Biomedical research, a tool to address the health issues that affect African popu-lations. 2013; 50.

21. Ramsay M, Tiemessen CT, Choudhury A, Soodyall H. Africa: the next frontier for human disease genediscovery? HumMol Genet. 2011; 20: R214–220. doi: 10.1093/hmg/ddr401 PMID: 21908518

22. Mulder NJ, Adebiyi E, Alami R, Benkahla A, Brandful J, Doumbia S, Everett D, Fadlelmola FM, GabounF, Gaseitsiwe S, Ghazal H, Hazelhurst S, HideW, Ibrahimi A, Jaufeerally Fakim Y, Jongeneel CV, Jou-bert F, Kassim S, Kayondo J, Kumuthini J, Lyantagaye S, Makani J, Mansour Alzohairy A, Masiga D,Moussa A, Nash O, Ouwe Missi Oukem-Boyer O, Owusu-Dabo E, Panji S, Patterton H, Radouani F,Sadki K, Seghrouchni F, Tastan Bishop Ö, Tiffin N, Ulenga N; H3ABioNet Consortium. H3ABioNet, asustainable pan-African bioinformatics network for human heredity and health in Africa.Genome Res.2015 Dec 1. [Epub ahead of print]

23. Bor J, Herbst AJ, Newell M- L, Bärnighausen T. Increases in adult life expectancy in rural South Africa:valuing the scale-up of HIV treatment. Science. 2013; 339: 961–965. doi: 10.1126/science.1230413PMID: 23430655

24. Shafer RW, Jung DR, Betts BJ. Human immunodeficiency virus type 1 reverse transcriptase and prote-ase mutation search engine for queries. Nat Med. 2000; 6: 1290–1292. doi: 10.1038/81407 PMID:11062545

25. De Oliveira T, Shafer RW, Seebregts C. Public database for HIV drug resistance in southern Africa.Nature. 2010; 464: 673. doi: 10.1038/464673c

26. Libin P, Beheydt G, Deforche K, Imbrechts S, Ferreira F, Van Laethem K, et al. RegaDB: community-driven data management and analysis for infectious diseases. Bioinforma Oxf Engl. 2013; 29: 1477–1480. doi: 10.1093/bioinformatics/btt162

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 14 / 15

27. Conradie F, Wilson D, Basson A, de Oliveira T, Hunt G, Joel D, et al. The 2012 southern African ARVdrug resistance testing guidelines. South Afr J HIV Med. 2012; 13: 162–167.

28. Lessells RJ, Avalos A, de Oliveira T. Implementing HIV-1 genotypic resistance testing in antiretroviraltherapy programs in Africa: needs, opportunities, and challenges. AIDS Rev. 2013; 15: 221–229.PMID: 24322382

29. Pepper MS. Launch of the Southern African Human Genome Programme. 2011; 287–288. PMID:21837864

30. Wright GE, Koornhof PG, Adeyemo AA, Tiffin N. Ethical and legal implications of whole genome andwhole exome sequencing in African populations. BMCMed Ethics. 2013; 21.

31. Olivier BG, Rohwer JM, Hofmeyr J-HS. Modelling cellular systems with PySCeS. Bioinforma Oxf Engl.2005; 21: 560–561. doi: 10.1093/bioinformatics/bti046

32. Olivier BG, Snoep JL. Web-based kinetic modelling using JWSOnline. Bioinforma Oxf Engl. 2004; 20:2143–2144. doi: 10.1093/bioinformatics/bth200

PLOS Computational Biology | DOI:10.1371/journal.pcbi.1004395 February 4, 2016 15 / 15